Research and development of an intrusion warning system using advanced artificial intelligence algorithms

Abstract

Given the extremely complicated problems of theft, having an intrusion warning system, especially at construction sites, is extremely urgent. In this study, we will build an automatic intrusion warning system when someone enters an area on a construction site automatically and accurately. Specifically, we used advanced deep learning models such as YOLOv5 and YOLOv8 to obtain the coordinates of the object and then compared them with the coordinates of the monitored area to determine whether the conduct was a violation. The results achieved in this study were very good when the YOLOv5n model achieved an average accuracy of more than 91% with a sensitivity of more than 84% and a processing speed of more than 12 frames per second, similar to the YOLOv8n model that achieved an average accuracy of more than 92%, with a sensitivity of more than 82% and a processing speed of more than 15 frames per second.